Re: Running utlrp creates too many Pnnn processes

From: Yong Huang <yong321_at_yahoo.com>
Date: Mon, 29 Dec 2008 15:44:53 -0800 (PST)
Message-ID: <918911.85961.qm@web80604.mail.mud.yahoo.com>


> I'll pursue changing the cpu_count parameter because, as you said,
> it will solve the majority of the problems.
> ...
> - Maureen

Why not precisely target the parameter you want to change instead of a "broad-spectrum" parameter? It's always a good idea to check dependent parameters when you change an unfamiliar one. Even though cpu_count seems to be modifiable with alter system (v$parameter.issys_modifiable is 'IMMEDIATE'), you need to bounce the instance to see the difference. My test on 10gR2 shows that it not only changes parallel_max_servers, but db_cache_size as well (and my sga_target is 0), and therefore many other parameters associated with the latter (18 in my count), including _small_table_threshold. In 11g, it also changes _sqlmon_max_plan. These could have quite noticeable performance impact.

Yong Huang       

--
http://www.freelists.org/webpage/oracle-l
Received on Mon Dec 29 2008 - 17:44:53 CST

Original text of this message